Biostatistics project: Randomized phase III study of panitumumab with FOLFOX4 for first-line treatment of metastatic colorectal cancer
Authors: Lucia Maninetti, Arianna Rigamonti, Michele Sala, Jacopo Tettamanti
ABSTRACT
Purpose. Panitumumab, a fully human anti–epidermal growth factor receptor (EGFR) monoclonal antibody that improves progression-free survival (PFS), is approved as monotherapy for patients with chemotherapy- refractory metastatic colorectal cancer (mCRC). The Panitumumab Randomized Trial in Combination With Chemotherapy for Metastatic Colorectal Cancer to Determine Efficacy (PRIME) was designed to evaluate the efficacy and safety of panitumumab plus infusional fluorouracil, leucovorin, and oxaliplatin (FOLFOX4) versus FOLFOX4 alone as initial treatment for mCRC.
Patients and Methods. In this multicenter, phase III trial, patients with no prior chemotherapy for mCRC, Eastern Cooperative Oncology Group performance status of 0 to 2, and available tissue for biomarker testing were randomly assigned 1:1 to receive panitumumab-FOLFOX4 versus FOLFOX4. The primary end point was PFS; overall survival (OS) was a secondary end point. Results were prospectively analyzed on an intent-to-treat basis by tumor KRAS status.
Results. KRAS results were available for 93% of the 1,183 patients randomly assigned. In the wild-type (WT) KRAS stratum, panitumumab-FOLFOX4 significantly improved PFS compared with FOLFOX4 (median PFS, 9.6 v 8.0 months, respectively; hazard ratio [HR], 0.80; 95% CI, 0.66 to 0.97; P 0.02). A nonsignificant increase in OS was also observed for panitumumab-FOLFOX4 versus FOLFOX4 (median OS, 23.9 v 19.7 months, respectively; HR, 0.83; 95% CI, 0.67 to 1.02; P 0.072). In the mutant KRAS stratum, PFS was significantly reduced in the panitumumab-FOLFOX4 arm versus the FOLFOX4 arm (HR, 1.29; 95% CI, 1.04 to 1.62; P 0.02), and median OS was 15.5 months versus 19.3 months, respectively (HR, 1.24; 95% CI, 0.98 to 1.57; P 0.068). Adverse event rates were generally comparable across arms with the exception of toxicities known to be associated with anti-EGFR therapy.
Conclusion. This study demonstrated that panitumumab-FOLFOX4 was well tolerated and significantly improved PFS in patients with WT KRAS tumors and underscores the importance of KRAS testing for patients with mCRC.
INTRODUCTION
Colorectal cancer (CRC) is the third most common cancer among men and women in the United States. Worldwide, there are more than one million new cases of CRC each year. On the basis of its role in the pathogenesis of CRC, the epidermal growth factor receptor (EGFR) has proven to be a clinically meaningful target for monoclonal antibodies (mAbs) with efficacy established in all lines of treatment of metastatic CRC (mCRC). Panitumumab is a fully human mAb targeting the EGFR. Retrospectively analyzed studies identified KRAS mutation in tumors as a negative predictive factor for panitumumab and cetuximab for improved response rate (RR), progression-free survival (PFS), and overall survival (OS). In September 2007, a prospectively defined, retrospective analysis of the pivotal phase III study of panitumumab as monotherapy in the mCRC setting provided evidence that clinical benefit was specific to patients with wild-type (WT) KRAS tumors. The Panitumumab Randomized Trial in Combination With Chemotherapy for Metastatic Colorectal Cancer to Determine Efficacy (PRIME) is an open-label, randomized, multicenter, phase III trial prospectively investigating panitumumab plus infusional fluorouracil, leucovorin, and oxaliplatin (FOLFOX4) versus FOLFOX4 alone as first-line treatment for mCRC in patients with WT KRAS tumors. Originally designed to compare the treatment effect in all randomly assigned patients, the trial was amended to focus on prospective hypothesis testing in the WT KRAS stratum.
PATIENTS AND METHODS
Study Design and Treatment Schedule. This is an open-label, multicenter, phase III trial that compared the efficacy of panitumumab-FOLFOX4 with FOLFOX4 alone in patients with previously untreated mCRC according to tumor KRAS status. Patients were randomly assigned 1:1 to receive either panitumumab-FOLFOX4 or FOLFOX4. Panitumumab was administered intravenously (IV) every 2 weeks on day 1 before FOLFOX4 chemotherapy.FOLFOX4 was administered every 2 weeks. Treatment was administered until progression or unacceptable toxicity. Objective tumor response was evaluated by blinded central radiology review using modified Response Evaluation Criteria in Solid Tumors (RECIST) in all patients with baseline measurable disease per central review. Patients were evaluated every 8 weeks until progression. Responses were confirmed at least 4 weeks later. Patients were observed for safety 30 days after the last study drug administration and for survival every 3 months. Adverse events (AEs) were graded using the Common Terminology Criteria for Adverse Events (version 3.0) with modifications for specific skin- and nail-related toxicities.
STATISTICS ANALYSIS
Import Libraries
library(readxl)
library(ggplot2)
library(ggpubr)
library(survival)
library(survminer)##
## Attaching package: 'survminer'## The following object is masked from 'package:survival':
##
## myelomalibrary(ggsurvfit)
library(dplyr)##
## Attaching package: 'dplyr'## The following objects are masked from 'package:stats':
##
## filter, lag## The following objects are masked from 'package:base':
##
## intersect, setdiff, setequal, unionlibrary(magrittr)
library(gridExtra)##
## Attaching package: 'gridExtra'## The following object is masked from 'package:dplyr':
##
## combineRead the Excel File and all the Sheets
Note: change the path with your own
colon1 <- read_excel("/Users/ariannarigamonti/Desktop/colon.xlsx", sheet = "adae_pds2019")
colon2 <- read_excel("/Users/ariannarigamonti/Desktop/colon.xlsx", sheet = "adlb_pds2019")
colon3 <- read_excel("/Users/ariannarigamonti/Desktop/colon.xlsx", sheet = "adls_pds2019")
colon4 <- read_excel("/Users/ariannarigamonti/Desktop/colon.xlsx", sheet = "adpm_pds2019")
colon5 <- read_excel("/Users/ariannarigamonti/Desktop/colon.xlsx", sheet = "adrsp_pds2019")
colon6 <- read_excel("/Users/ariannarigamonti/Desktop/colon.xlsx", sheet = "adsl_pds2019")
colon7 <- read_excel("/Users/ariannarigamonti/Desktop/colon.xlsx", sheet ="biomark_pds2019")Merge the Interesting Variables in One Final Dataframe
col1_col6_merged <- merge(colon1, colon6, by.x = "SUBJID")
col1_col6_col7_merged <- merge(col1_col6_merged, colon7, by.x = "SUBJID")
col6_col7_merged <- merge(colon6, colon7, by.x = "SUBJID")
df_final_pre <- unique(col1_col6_col7_merged)
df_final <- df_final_pre %>%
group_by(SUBJID, AEPT) %>%
arrange(desc(AESEVCD)) %>%
slice(1) %>%
ungroup()
df_final1 <- df_final_pre %>%
group_by(SUBJID) %>%
arrange(desc(AESEVCD)) %>%
slice(1) %>%
ungroup()Divide Patients in Mutant (mt) and Wild-Type (wt) for KRAS (Exon 2) Gene ‘BMMTR1’
mutant <- subset(col6_col7_merged, BMMTR1 == "Mutant" | BMMTR2 == "Mutant" | BMMTR3 == "Mutant" | BMMTR4 == "Mutant" |
BMMTR5 == "Mutant" | BMMTR6 == "Mutant" | BMMTR7 == "Mutant" | BMMTR15 == "Mutant" | BMMTR16 == "Mutant")
wild_type <- subset(col6_col7_merged,
BMMTR1 == "Wild-type" &
BMMTR2 == "Wild-type" &
BMMTR3 == "Wild-type" &
BMMTR4 == "Wild-type" &
BMMTR5 == "Wild-type" &
BMMTR6 == "Wild-type" &
BMMTR7 == "Wild-type" &
BMMTR15 == "Wild-type" &
BMMTR16 == "Wild-type")Divide patients based on AE grade (1-2 and 3-4)
skin_ae <- subset(df_final1, AESOC == "Skin and subcutaneous tissue disorders")
df_grade_1_2 <- skin_ae[skin_ae$AESEVCD %in% c(1, 2), ]
df_grade_3_4 <- skin_ae[skin_ae$AESEVCD %in% c(3, 4), ]DESCRIPTIVE STATISTICS
Frequencies Barplot of MT and WT Patients
mutant1 <- mutant
mutant1$BMMTR = "Mutant"
wild_type1 <- wild_type
wild_type1$BMMTR = "Wild-type"
df_mt_wt <- rbind(mutant1, wild_type1)
df_final1 <- merge(df_final1, df_mt_wt[, c("SUBJID", "BMMTR")], by = "SUBJID", all.x = TRUE)
freq_mt_wt <- table(df_mt_wt$BMMTR)
barplot(freq_mt_wt,
main = "KRAS type",
ylab = "Frequency",
col = c("salmon", "yellowgreen"),
ylim = c(0,500))
# Creare un grafico a barre con ggplot2
ggplot(df_mt_wt, aes(x = BMMTR, fill = BMMTR)) +
geom_bar() +
labs(title = "KRAS Type", x = "KRAS Type", y = "Frequency") +
scale_fill_manual(values = c("salmon", "yellowgreen")) +
theme_minimal()
# Calcolare le percentuali
df_mt_wt_percent <- df_mt_wt %>%
count(BMMTR) %>%
mutate(Percent = n / sum(n) * 100)
# Creare un grafico a barre in percentuale con ggplot2
ggplot(df_mt_wt_percent, aes(x = BMMTR, y = Percent, fill = BMMTR)) +
geom_bar(stat = "identity") +
labs(title = "KRAS Type distribution", x = "KRAS Type", y = "Percentage") +
scale_fill_manual(values = c("salmon", "yellowgreen")) +
theme_minimal() +
geom_text(aes(label = paste0(round(Percent, 1), "%")), vjust = -0.5) +
theme(plot.title = element_text(hjust = 0.5))
Adverse events
# Adverse events on all patients (regarless the therapy)
freq_AESOC <- round(((table(df_final$AESOC)) / nrow(df_final)) * 100, 1)
# Crea un dataframe con le frequenze percentuali
df_pie <- data.frame(categories = names(freq_AESOC), frequencies = freq_AESOC)
# Crea la pie chart con ggplot2
ggplot(df_pie, aes(x = "", y = freq_AESOC, fill = categories)) +
geom_bar(stat = "identity", width = 1) +
coord_polar("y", start = 0) + # Imposta il tipo di coordinate su polare
labs(title = "Therapy Adverse Events Categories",
fill = "Categories",
x = NULL,
y = "Percentage") + # Rimuovi l'etichetta dell'asse x
theme_minimal() + # Stile del tema
scale_fill_manual(values = rainbow(length(freq_AESOC))) # Imposta i colori manualmente## Don't know how to automatically pick scale for object of type <table>.
## Defaulting to continuous.
# Adverse events based on the therapy
# FOLFOX4 alone
folfox <- subset(df_final, TRT == "FOLFOX alone")
freq_AESOC_folfox <- round(((table(folfox$AESOC)) / nrow(folfox)) * 100, 1)
# Crea un dataframe con le frequenze percentuali
df_pie_folfox <- data.frame(categories = names(freq_AESOC_folfox), frequencies = freq_AESOC_folfox)
# Crea la pie chart con ggplot2
ggplot(df_pie_folfox, aes(x = "", y = freq_AESOC_folfox, fill = categories)) +
geom_bar(stat = "identity", width = 1) +
coord_polar("y", start = 0) + # Imposta il tipo di coordinate su polare
labs(title = "Folfox Adverse Events",
fill = "Categories",
x = NULL,
y = "Percentage") + # Rimuovi l'etichetta dell'asse x
theme_minimal() + # Stile del tema
scale_fill_manual(values = rainbow(length(freq_AESOC_folfox))) # Imposta i colori manualmente## Don't know how to automatically pick scale for object of type <table>.
## Defaulting to continuous.
# FOLFOX4 alone
folfox_panitumumab <- subset(df_final, TRT == "Panitumumab + FOLFOX")
freq_AESOC_folfox_panitumumab <- round(((table(folfox_panitumumab$AESOC)) / nrow(folfox_panitumumab)) * 100, 1)
# Crea un dataframe con le frequenze percentuali
df_pie_folfox_panitumumab <- data.frame(categories = names(freq_AESOC_folfox_panitumumab), frequencies = freq_AESOC_folfox_panitumumab)
# Crea la pie chart con ggplot2
ggplot(df_pie_folfox_panitumumab, aes(x = "", y = freq_AESOC_folfox_panitumumab, fill = categories)) +
geom_bar(stat = "identity", width = 1) +
coord_polar("y", start = 0) + # Imposta il tipo di coordinate su polare
labs(title = "Folfox + Panitumumab Adverse Events",
fill = "Categories",
x = NULL,
y = "Percentage") + # Rimuovi l'etichetta dell'asse x
theme_minimal() + # Stile del tema
scale_fill_manual(values = rainbow(length(freq_AESOC_folfox_panitumumab))) # Imposta i colori manualmente## Don't know how to automatically pick scale for object of type <table>.
## Defaulting to continuous.
Frequencies Barplot of Adverse Events Grade
freq_AESEVCD <- table(df_final$AESEVCD)
df_bar <- data.frame(Grade = as.numeric(names(freq_AESEVCD)), Frequency = freq_AESEVCD)
# Crea il barplot con ggplot2
ggplot(df_bar, aes(x = factor(Grade), y = freq_AESEVCD, fill = factor(Grade))) +
geom_bar(stat = "identity") +
labs(title = "Adverse Events Grade Distribution",
x = "Grade (1-4)",
y = "Frequency") +
scale_fill_manual(name = 'Grade', values = c("chartreuse1", "yellow", "orange", "red")) + # Colori delle barre
ylim(0, 1000) + # Imposta i limiti sull'asse y
theme_minimal() +
theme(axis.text.x = element_text(angle = 0, hjust = 1))
SURVIVAL ANALYSIS
The primary objective of this study was to assess the treatment effect of the addition of panitumumab to FOLFOX4 on PFS as initial therapy for mCRC in patients with WT KRAS tumors and also in patients with MT KRAS tumors. The study was amended to compare PFS (primary end point) with OS (secondary end point) according to KRAS status before any efficacy analyses were carried out.
Treatment effects on PFS and OS were estimated using stratified Cox proportional hazards models and the Kaplan–Meier method. An exact 95% confidence interval (CI) was estimated for a stratified odds ratio for objective response rate (ORR). The randomization stratification factors were used for the stratified analyses.
# Calcolare la mediana del tempo di sopravvivenza per tutti i pazienti (PFS)
fit_pfs <- survfit(Surv(PFSDYCR, PFSCR) ~ TRT, data = df_mt_wt)
summary(fit_pfs)$table## records n.max n.start events rmean se(rmean) median
## TRT=FOLFOX alone 411 411 411 361 346.0519 15.01651 273
## TRT=Panitumumab + FOLFOX 423 423 423 366 365.2060 16.04395 270
## 0.95LCL 0.95UCL
## TRT=FOLFOX alone 238 290
## TRT=Panitumumab + FOLFOX 231 285median_pfs <- summary(fit_pfs)$table
# Convertire la mediana da giorni a mesi
median_pfs_months <- round(median_pfs[,"median"] / 30.44, 1)
# Calcolare la mediana del tempo di sopravvivenza per tutti i pazienti (OS)
fit_os <- survfit(Surv(DTHDY, DTH) ~ TRT, data = df_mt_wt)
median_os <- summary(fit_os)$table
summary(fit_os)$table## records n.max n.start events rmean se(rmean) median
## TRT=FOLFOX alone 411 411 411 310 696.3368 21.81274 600
## TRT=Panitumumab + FOLFOX 423 423 423 307 695.3658 22.76780 602
## 0.95LCL 0.95UCL
## TRT=FOLFOX alone 536 673
## TRT=Panitumumab + FOLFOX 546 672# Convertire la mediana da giorni a mesi
median_os_months <- round(median_os[,"median"] / 30.44, 1)
# Eseguire il log-rank test per PFS
survdiff_pfs <- survdiff(Surv(PFSDYCR, PFSCR) ~ TRT, data = df_mt_wt)
print(survdiff_pfs)## Call:
## survdiff(formula = Surv(PFSDYCR, PFSCR) ~ TRT, data = df_mt_wt)
##
## N Observed Expected (O-E)^2/E (O-E)^2/V
## TRT=FOLFOX alone 411 361 351 0.282 0.55
## TRT=Panitumumab + FOLFOX 423 366 376 0.264 0.55
##
## Chisq= 0.6 on 1 degrees of freedom, p= 0.5# Eseguire il log-rank test per OS
survdiff_os <- survdiff(Surv(DTHDY, DTH) ~ TRT, data = df_mt_wt)
print(survdiff_os)## Call:
## survdiff(formula = Surv(DTHDY, DTH) ~ TRT, data = df_mt_wt)
##
## N Observed Expected (O-E)^2/E (O-E)^2/V
## TRT=FOLFOX alone 411 310 305 0.0698 0.139
## TRT=Panitumumab + FOLFOX 423 307 312 0.0684 0.139
##
## Chisq= 0.1 on 1 degrees of freedom, p= 0.7# Visualizzare la curva di sopravvivenza con ggsurvfit per PFS
p1 <- ggsurvplot(fit_pfs, data = df_mt_wt, pval = TRUE, risk.table = TRUE,
title = "Kaplan-Meier Curve for mCRC PFS by treatment",
xlab = "Follow-up time (days)", ylab = "Progression Free Survival (PFS)",
palette = c("blue", "gold2"),
conf.int = TRUE,
legend.labs = c("FOLFOX4", "Panitumumab + FOLFOX4"),
legend.title = "Treatment")
# Aggiungi la mediana al plot PFS
p1$plot <- p1$plot +
geom_vline(xintercept = median_pfs[,"median"], linetype = "dashed", color = "red")
# Visualizzare la curva di sopravvivenza con ggsurvfit per OS
p2 <- ggsurvplot(fit_os, data = df_mt_wt, pval = TRUE, risk.table = TRUE,
title = "Kaplan-Meier Curve for mCRC OS by treatment",
xlab = "Follow-up time (days)", ylab = "Overall Survival (OS)",
palette = c("blue", "gold2"),
conf.int = TRUE,
legend.labs = c("FOLFOX4", "Panitumumab + FOLFOX4"),
legend.title = "Treatment")
# Aggiungi la mediana al plot OS
p2$plot <- p2$plot +
geom_vline(xintercept = median_os[,"median"], linetype = "dashed", color = "red")
# Creare una tabella con le mediane di sopravvivenza in mesi
median_table <- data.frame(
Treatment = rep(c("FOLFOX4", "Panitumumab + FOLFOX4"), 2),
Survival_Type = c("PFS", "PFS", "OS", "OS"),
Median_Survival_Months = c(median_pfs_months, median_os_months)
)
# Creare un grafico della tabella delle mediane
table_plot <- tableGrob(median_table, rows = NULL)
# Disporre i plot in un unico layout con la tabella a lato
grid.arrange(p1$plot, p2$plot, table_plot, ncol = 1, nrow = 3, heights = c(2, 2, 1))
# Creare la curva di sopravvivenza di Kaplan-Meier per PFS (wild-type)
fit_pfs <- survfit(Surv(PFSDYCR, PFSCR) ~ TRT, data = wild_type1)
summary(fit_pfs)$table## records n.max n.start events rmean se(rmean) median
## TRT=FOLFOX alone 169 169 169 145 351.7932 22.0711 285
## TRT=Panitumumab + FOLFOX 181 181 181 142 479.9806 28.9404 341
## 0.95LCL 0.95UCL
## TRT=FOLFOX alone 237 337
## TRT=Panitumumab + FOLFOX 287 424median_pfs <- summary(fit_pfs)$table
median_pfs## records n.max n.start events rmean se(rmean) median
## TRT=FOLFOX alone 169 169 169 145 351.7932 22.0711 285
## TRT=Panitumumab + FOLFOX 181 181 181 142 479.9806 28.9404 341
## 0.95LCL 0.95UCL
## TRT=FOLFOX alone 237 337
## TRT=Panitumumab + FOLFOX 287 424median_pfs_months <- round(median_pfs[,"median"] / 30.44, 1)
median_pfs_months## TRT=FOLFOX alone TRT=Panitumumab + FOLFOX
## 9.4 11.2# Visualizzare la curva di sopravvivenza con ggsurvfit
p1 <- ggsurvplot(fit_pfs, data = wild_type1, pval = TRUE, risk.table = TRUE,
title = "Kaplan-Meier Curve for mCRC KRAS WT PFS by treatment",
xlab = "Follow-up time (days)", ylab = "Progression Free Survival (PFS)",
palette = c("blue", "gold2"),
conf.int = TRUE,
legend.labs = c("FOLFOX4", "Panitumumab + FOLFOX4"),
legend.title = "Treatment")
# Aggiungi la mediana al plot PFS
p1$plot <- p1$plot +
geom_vline(xintercept = median_pfs[,"median"], linetype = "dashed", color = "red")
# Creare la curva di sopravvivenza di Kaplan-Meier per PFS (mutant)
fit_pfs_mutant <- survfit(Surv(PFSDYCR, PFSCR) ~ TRT, data = mutant1)
median_pfs_mutant <- summary(fit_pfs_mutant)$table
median_pfs_mutant_months <- round(median_pfs_mutant[,"median"] / 30.44, 1)
# Visualizzare la curva di sopravvivenza con ggsurvfit
p2 <- ggsurvplot(fit_pfs_mutant, data = mutant1, pval = TRUE, risk.table = TRUE,
title = "Kaplan-Meier Curve for mCRC KRAS MT PFS by treatment",
xlab = "Follow-up time (days)", ylab = "Progression Free Survival (PFS)",
palette = c("blue", "gold2"),
conf.int = TRUE,
legend.labs = c("FOLFOX4","Panitumumab + FOLFOX4"),
legend.title = "Treatment")
# Aggiungi la mediana al plot OS
p2$plot <- p2$plot +
geom_vline(xintercept = median_pfs_mutant[,"median"], linetype = "dashed", color = "red")
# Creare la curva di sopravvivenza di Kaplan-Meier per OS (wild-type)
fit_os <- survfit(Surv(DTHDY, DTH) ~ TRT, data = wild_type1)
median_os <- summary(fit_os)$table
median_os_months <- round(median_os[,"median"] / 30.44, 1)
# Visualizzare la curva di sopravvivenza con ggsurvfit
p3 <- ggsurvplot(fit_os, data = wild_type1, pval = TRUE, risk.table = TRUE,
title = "Kaplan-Meier Curve for mCRC KRAS WT OS by treatment",
xlab = "Follow-up time (days)", ylab = "Overall Survival (OS)",
palette = c("blue", "gold2"),
conf.int = TRUE,
legend.labs = c("FOLFOX4", "Panitumumab + FOLFOX4"),
legend.title = "Treatment")
# Aggiungi la mediana al plot OS
p3$plot <- p3$plot +
geom_vline(xintercept = median_os[,"median"], linetype = "dashed", color = "red")
# Creare la curva di sopravvivenza di Kaplan-Meier per OS (mutant)
fit_os_mutant <- survfit(Surv(DTHDY, DTH) ~ TRT, data = mutant1)
median_os_mutant <- summary(fit_os_mutant)$table
median_os_mutant_months <- round(median_os_mutant[,"median"] / 30.44, 1)
# Visualizzare la curva di sopravvivenza con ggsurvfit
p4 <- ggsurvplot(fit_os_mutant, data = mutant1, pval = TRUE, risk.table = TRUE,
title = "Kaplan-Meier Curve for mCRC KRAS MT OS by treatment",
xlab = "Follow-up time (days)", ylab = "Overall Survival (OS)",
palette = c("blue", "gold2"),
conf.int = TRUE,
legend.labs = c("FOLFOX4", "Panitumumab + FOLFOX4"),
legend.title = "Treatment")
# Aggiungi la mediana al plot OS
p4$plot <- p4$plot +
geom_vline(xintercept = median_os_mutant[,"median"], linetype = "dashed", color = "red")
# Creare una tabella con le mediane di sopravvivenza
median_table <- data.frame(
Treatment = rep(c("FOLFOX4", "Panitumumab + FOLFOX4"), 4),
Survival_Type = c("PFS-WT", "PFS-WT", "PFS-MT", "PFS-MT", "OS-WT", "OS-WT", "OS-MT", "OS-MT"),
Median_Survival_Months = c(median_pfs_months, median_pfs_mutant_months, median_os_months, median_os_mutant_months)
)
# Creare un grafico della tabella delle mediane
table_plot <- tableGrob(median_table, rows = NULL)
# Disporre i plot in un unico layout
grid.arrange(p1$plot, p2$plot, p3$plot, p4$plot, ncol = 2, nrow = 2)
grid.arrange(table_plot, ncol=1, nrow = 1)
Skin toxicity was defined as any treatment-emergent adverse event indicative of a skin disorder, representing a composite category of adverse event terms including but not limited to rash, dermatitis acneiform, pruritus, dry skin, skin fissures, and erythema. Retrospective post-hoc analyses were carried out to determine the effect of skin toxicity on efficacy end points, in- cluding PFS (by central review) and OS. A stratified Cox proportional hazards model was used to examine the relationship between worst grade skin toxicity severity (grade 2–4: grade 0–1) and PFS/OS. ORR by central review and worst grade skin toxicity was provided.
# Creare una variabile AE per indicare il grado di AE
df_final1$AE <- ifelse(df_final1$AESEVCD %in% c(1, 2), "Low", "High")
# Creare la curva di sopravvivenza di Kaplan-Meier per PFS in base al grado di AE
fit_pfs_skin <- survfit(Surv(PFSDYCR, PFSCR) ~ AE, data = df_final1)
summary(fit_pfs_skin)$table## records n.max n.start events rmean se(rmean) median 0.95LCL 0.95UCL
## AE=High 175 175 175 145 420.9534 28.19621 287 268 328
## AE=Low 433 433 433 383 353.1835 14.66345 275 234 290median_pfs_skin <- summary(fit_pfs_skin)$table
median_pfs_months_skin <- round(median_pfs_skin[,"median"] / 30.44, 1)
# Visualizzare la curva di sopravvivenza con ggsurvfit
p1 <- ggsurvplot(fit_pfs_skin, data = df_final1, pval = TRUE, risk.table = TRUE,
title = "Kaplan-Meier Curve for mCRC PFS by AE grade",
xlab = "Follow-up time (days)", ylab = "Progression Free Survival (PFS)",
palette = c("blue", "gold2"),
conf.int = TRUE,
legend.labs = c("High", "Low"),
legend.title = "Adverse Events Grade")
# Aggiungi la mediana al plot PFS
p1$plot <- p1$plot +
geom_vline(xintercept = median_pfs_skin[,"median"], linetype = "dashed", color = "red")
# Creare la curva di sopravvivenza di Kaplan-Meier per OS in base al grado di AE
fit_os_skin <- survfit(Surv(DTHDY, DTH) ~ AE, data = df_final1)
median_os_skin <- summary(fit_os_skin)$table
median_os_months_skin <- round(median_os_skin[,"median"] / 30.44, 1)
# Visualizzare la curva di sopravvivenza con ggsurvfit
p2 <- ggsurvplot(fit_os_skin, data = df_final1, pval = TRUE, risk.table = TRUE,
title = "Kaplan-Meier Curve for mCRC OS by AE grade",
xlab = "Follow-up time (days)", ylab = "Overall Survival (OS)",
palette = c("blue", "gold2"),
conf.int = TRUE,
legend.labs = c("High", "Low"),
legend.title = "Adverse Events Grade")
# Aggiungi la mediana al plot OS
p2$plot <- p2$plot +
geom_vline(xintercept = median_os_skin[,"median"], linetype = "dashed", color = "red")
# Creare una tabella con le mediane di sopravvivenza
median_table_skin <- data.frame(
AE_Grade = rep(c("High", "Low"), 2),
Survival_Type = c("PFS", "PFS", "OS", "OS"),
Median_Survival_Months = c(median_pfs_months_skin, median_os_months_skin)
)
# Creare un grafico della tabella delle mediane
table_plot <- tableGrob(median_table_skin, rows = NULL)
# Disporre i plot in un unico layout
grid.arrange(p1$plot, p2$plot, ncol = 1)
grid.arrange(table_plot, ncol = 1, nrow = 1)
# Creare la curva di sopravvivenza di Kaplan-Meier per PFS in base a ECOC
fit_pfs_ecog <- survfit(Surv(PFSDYCR, PFSCR) ~ B_ECOG, data = df_final1)
median_pfs_ecog <- summary(fit_pfs_ecog)$table
median_pfs_months_ecog <- round(median_pfs_ecog[,"median"] / 30.44, 1)
# Visualizzare la curva di sopravvivenza con ggsurvfit
p1 <- ggsurvplot(fit_pfs_ecog, data = df_final1, pval = TRUE, risk.table = TRUE,
title = "Kaplan-Meier Curve for mCRC PFS by Baseline ECOG",
xlab = "Follow-up time (days)", ylab = "Progression Free Survival (PFS)",
palette = c("blue", "gold2", "olivedrab4"),
conf.int = TRUE,
legend.labs = c("Fully Active", "In bed less than 50% of the time", "Symptoms but ambulatory"),
legend.title = "Baseline ECOG")
# Aggiungi la mediana al plot PFS
p1$plot <- p1$plot +
geom_vline(xintercept = median_pfs_ecog[,"median"], linetype = "dashed", color = "red")
# Creare la curva di sopravvivenza di Kaplan-Meier per OS in base a ECOG
fit_os_ecog <- survfit(Surv(DTHDY, DTH) ~ B_ECOG, data = df_final1)
median_os_ecog <- summary(fit_os_ecog)$table
median_os_months_ecog <- round(median_os_ecog[,"median"] / 30.44, 1)
# Visualizzare la curva di sopravvivenza con ggsurvfit
p2 <- ggsurvplot(fit_os_ecog, data = df_final1, pval = TRUE, risk.table = TRUE,
title = "Kaplan-Meier Curve for mCRC OS by Baseline ECOG",
xlab = "Follow-up time (days)", ylab = "Overall Survival (OS)",
palette = c("blue", "gold2", "olivedrab4"),
conf.int = TRUE,
legend.labs = c("Fully Active", "In bed less than 50% of the time", "Symptoms but ambulatory"),
legend.title = "Baseline ECOG")
# Aggiungi la mediana al plot OS
p2$plot <- p2$plot +
geom_vline(xintercept = median_os_ecog[,"median"], linetype = "dashed", color = "red")
# Creare una tabella con le mediane di sopravvivenza
median_table_ecog <- data.frame(
ECOG = rep(c("Fully Active", "In bed less than 50% of the time", "Symptoms but ambulatory"), 2),
Survival_Type = c("PFS", "PFS","PFS","OS", "OS", "OS"),
Median_Survival_Months = c(median_pfs_months_ecog, median_os_months_ecog)
)
# Creare un grafico della tabella delle mediane
table_plot <- tableGrob(median_table_ecog, rows = NULL)
# Disporre i plot in un unico layout
grid.arrange(p1$plot, p2$plot, ncol = 1)
grid.arrange(table_plot, ncol = 1, nrow = 1)
hist(df_final1$AGE, xlab='Age [years]', main='Age distribution', col = "skyblue")
summary(df_final1$AGE)## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 27.00 54.00 61.00 60.74 69.00 83.00df_final1$agecat61 <- cut(df_final1$AGE, breaks=c(0, 55, Inf), labels=c("young", "old"))
head(df_final1)## SUBJID AEPT AESOC AESTDYI
## 1 3 Dermatitis acneiform Skin and subcutaneous tissue disorders 12
## 2 7 Folliculitis Infections and infestations 245
## 3 9 Exfoliative rash Skin and subcutaneous tissue disorders 3
## 4 13 Rash Skin and subcutaneous tissue disorders 37
## 5 15 Dry skin Skin and subcutaneous tissue disorders 72
## 6 17 Urticaria Skin and subcutaneous tissue disorders 175
## AEENDYI AESEVCD AECONT TRT ATRT PRSURG DTHDY
## 1 29 1 N Panitumumab + FOLFOX Panitumumab + FOLFOX Y 301
## 2 252 3 N Panitumumab + FOLFOX Panitumumab + FOLFOX Y 339
## 3 14 2 N Panitumumab + FOLFOX Panitumumab + FOLFOX Y 696
## 4 55 2 N Panitumumab + FOLFOX Panitumumab + FOLFOX Y 197
## 5 157 2 N Panitumumab + FOLFOX Panitumumab + FOLFOX Y 414
## 6 175 2 N FOLFOX alone FOLFOX alone Y 1043
## DTH PFSDYCR PFSCR LIVERMET DIAGMONS AGE SEX B_WEIGHT B_HEIGHT
## 1 1 54 1 Y 1.86885245901639E+16 65 Male 67 165
## 2 1 339 1 Y 2.78688524590163E+16 52 Female 53 160
## 3 1 301 1 Y 7570491803278680 42 Female 69 148
## 4 1 73 1 Y 2.0327868852459E+16 54 Male 62 160
## 5 1 414 1 Y 380327868852459 67 Female 48.8 155
## 6 0 226 0 Y 3.14754098360655E+16 57 Female 63.6 149
## RACE B_ECOG HISSUBTY B_METANM DIAGTYPE
## 1 White or Caucasian Symptoms but ambulatory No sub-type 5 Colon
## 2 Hispanic or Latino Symptoms but ambulatory Mucinous 3 Colon
## 3 Hispanic or Latino Fully active No sub-type 4 Colon
## 4 White or Caucasian Fully active Other 6 Rectal
## 5 White or Caucasian Symptoms but ambulatory No sub-type 2 Rectal
## 6 White or Caucasian Fully active No sub-type 2 Colon
## BMMTNM1 BMMTR1 BMMTNM2 BMMTR2
## 1 KRAS exon 2 (c12/13) Mutant KRAS exon 3 (c61) <NA>
## 2 KRAS exon 2 (c12/13) Mutant KRAS exon 3 (c61) <NA>
## 3 KRAS exon 2 (c12/13) Failure KRAS exon 3 (c61) <NA>
## 4 KRAS exon 2 (c12/13) Wild-type KRAS exon 3 (c61) Wild-type
## 5 KRAS exon 2 (c12/13) Failure KRAS exon 3 (c61) <NA>
## 6 KRAS exon 2 (c12/13) Mutant KRAS exon 3 (c61) <NA>
## BMMTNM3 BMMTR3 BMMTNM4 BMMTR4
## 1 KRAS exon 4 (c117/146) <NA> NRAS exon 2 (c12/13) <NA>
## 2 KRAS exon 4 (c117/146) <NA> NRAS exon 2 (c12/13) <NA>
## 3 KRAS exon 4 (c117/146) <NA> NRAS exon 2 (c12/13) <NA>
## 4 KRAS exon 4 (c117/146) Mutant NRAS exon 2 (c12/13) Wild-type
## 5 KRAS exon 4 (c117/146) <NA> NRAS exon 2 (c12/13) <NA>
## 6 KRAS exon 4 (c117/146) <NA> NRAS exon 2 (c12/13) <NA>
## BMMTNM5 BMMTR5 BMMTNM6 BMMTR6
## 1 NRAS exon 3 (c61) <NA> NRAS exon 4 (c117/146) <NA>
## 2 NRAS exon 3 (c61) <NA> NRAS exon 4 (c117/146) <NA>
## 3 NRAS exon 3 (c61) <NA> NRAS exon 4 (c117/146) <NA>
## 4 NRAS exon 3 (c61) Wild-type NRAS exon 4 (c117/146) Wild-type
## 5 NRAS exon 3 (c61) <NA> NRAS exon 4 (c117/146) <NA>
## 6 NRAS exon 3 (c61) <NA> NRAS exon 4 (c117/146) <NA>
## BMMTNM7 BMMTR7 BMMTNM15 BMMTR15
## 1 BRAF exon 15 (c600) <NA> KRAS exon 3 (c59/c61) <NA>
## 2 BRAF exon 15 (c600) <NA> KRAS exon 3 (c59/c61) <NA>
## 3 BRAF exon 15 (c600) <NA> KRAS exon 3 (c59/c61) <NA>
## 4 BRAF exon 15 (c600) Wild-type KRAS exon 3 (c59/c61) Wild-type
## 5 BRAF exon 15 (c600) <NA> KRAS exon 3 (c59/c61) <NA>
## 6 BRAF exon 15 (c600) <NA> KRAS exon 3 (c59/c61) <NA>
## BMMTNM16 BMMTR16 BMMTR AE agecat61
## 1 NRAS exon 3 (c59/c61) <NA> Mutant Low old
## 2 NRAS exon 3 (c59/c61) <NA> Mutant High young
## 3 NRAS exon 3 (c59/c61) <NA> <NA> Low young
## 4 NRAS exon 3 (c59/c61) Wild-type Mutant Low young
## 5 NRAS exon 3 (c59/c61) <NA> <NA> Low old
## 6 NRAS exon 3 (c59/c61) <NA> Mutant Low oldfit.age <- survfit(Surv(PFSDYCR, PFSCR) ~ agecat61, data=df_final1)
print(fit.age)## Call: survfit(formula = Surv(PFSDYCR, PFSCR) ~ agecat61, data = df_final1)
##
## n events median 0.95LCL 0.95UCL
## agecat61=young 178 152 285 234 332
## agecat61=old 430 376 280 238 290survfit2(Surv(PFSDYCR, PFSCR) ~ agecat61, data=df_final1) %>%
ggsurvfit(linewidth = 1) +
add_confidence_interval() + # add confidence interval
add_risktable() + # Add risk table
add_quantile(y_value = 0.5, color = "gray50", linewidth = 0.75)+ # Specify median survival
labs(title = "Kaplan-Meier Curve by age group",
x="Follow-up time (days)")+
scale_x_continuous(breaks = seq(0,1000,by=90))+
scale_color_discrete("Age group")+
scale_fill_discrete("Age group")+
add_pvalue("annotation",x=500)
df_final1$BMI <- with(df_final1, (as.numeric(B_WEIGHT) / as.numeric(B_HEIGHT)^2) * 10000)
df_final1$BMI_category <- cut(df_final1$BMI, breaks=c(0, 18.5, 30, Inf), labels=c("underweight", "normal weight", "overweight/obese"))
head(df_final1)## SUBJID AEPT AESOC AESTDYI
## 1 3 Dermatitis acneiform Skin and subcutaneous tissue disorders 12
## 2 7 Folliculitis Infections and infestations 245
## 3 9 Exfoliative rash Skin and subcutaneous tissue disorders 3
## 4 13 Rash Skin and subcutaneous tissue disorders 37
## 5 15 Dry skin Skin and subcutaneous tissue disorders 72
## 6 17 Urticaria Skin and subcutaneous tissue disorders 175
## AEENDYI AESEVCD AECONT TRT ATRT PRSURG DTHDY
## 1 29 1 N Panitumumab + FOLFOX Panitumumab + FOLFOX Y 301
## 2 252 3 N Panitumumab + FOLFOX Panitumumab + FOLFOX Y 339
## 3 14 2 N Panitumumab + FOLFOX Panitumumab + FOLFOX Y 696
## 4 55 2 N Panitumumab + FOLFOX Panitumumab + FOLFOX Y 197
## 5 157 2 N Panitumumab + FOLFOX Panitumumab + FOLFOX Y 414
## 6 175 2 N FOLFOX alone FOLFOX alone Y 1043
## DTH PFSDYCR PFSCR LIVERMET DIAGMONS AGE SEX B_WEIGHT B_HEIGHT
## 1 1 54 1 Y 1.86885245901639E+16 65 Male 67 165
## 2 1 339 1 Y 2.78688524590163E+16 52 Female 53 160
## 3 1 301 1 Y 7570491803278680 42 Female 69 148
## 4 1 73 1 Y 2.0327868852459E+16 54 Male 62 160
## 5 1 414 1 Y 380327868852459 67 Female 48.8 155
## 6 0 226 0 Y 3.14754098360655E+16 57 Female 63.6 149
## RACE B_ECOG HISSUBTY B_METANM DIAGTYPE
## 1 White or Caucasian Symptoms but ambulatory No sub-type 5 Colon
## 2 Hispanic or Latino Symptoms but ambulatory Mucinous 3 Colon
## 3 Hispanic or Latino Fully active No sub-type 4 Colon
## 4 White or Caucasian Fully active Other 6 Rectal
## 5 White or Caucasian Symptoms but ambulatory No sub-type 2 Rectal
## 6 White or Caucasian Fully active No sub-type 2 Colon
## BMMTNM1 BMMTR1 BMMTNM2 BMMTR2
## 1 KRAS exon 2 (c12/13) Mutant KRAS exon 3 (c61) <NA>
## 2 KRAS exon 2 (c12/13) Mutant KRAS exon 3 (c61) <NA>
## 3 KRAS exon 2 (c12/13) Failure KRAS exon 3 (c61) <NA>
## 4 KRAS exon 2 (c12/13) Wild-type KRAS exon 3 (c61) Wild-type
## 5 KRAS exon 2 (c12/13) Failure KRAS exon 3 (c61) <NA>
## 6 KRAS exon 2 (c12/13) Mutant KRAS exon 3 (c61) <NA>
## BMMTNM3 BMMTR3 BMMTNM4 BMMTR4
## 1 KRAS exon 4 (c117/146) <NA> NRAS exon 2 (c12/13) <NA>
## 2 KRAS exon 4 (c117/146) <NA> NRAS exon 2 (c12/13) <NA>
## 3 KRAS exon 4 (c117/146) <NA> NRAS exon 2 (c12/13) <NA>
## 4 KRAS exon 4 (c117/146) Mutant NRAS exon 2 (c12/13) Wild-type
## 5 KRAS exon 4 (c117/146) <NA> NRAS exon 2 (c12/13) <NA>
## 6 KRAS exon 4 (c117/146) <NA> NRAS exon 2 (c12/13) <NA>
## BMMTNM5 BMMTR5 BMMTNM6 BMMTR6
## 1 NRAS exon 3 (c61) <NA> NRAS exon 4 (c117/146) <NA>
## 2 NRAS exon 3 (c61) <NA> NRAS exon 4 (c117/146) <NA>
## 3 NRAS exon 3 (c61) <NA> NRAS exon 4 (c117/146) <NA>
## 4 NRAS exon 3 (c61) Wild-type NRAS exon 4 (c117/146) Wild-type
## 5 NRAS exon 3 (c61) <NA> NRAS exon 4 (c117/146) <NA>
## 6 NRAS exon 3 (c61) <NA> NRAS exon 4 (c117/146) <NA>
## BMMTNM7 BMMTR7 BMMTNM15 BMMTR15
## 1 BRAF exon 15 (c600) <NA> KRAS exon 3 (c59/c61) <NA>
## 2 BRAF exon 15 (c600) <NA> KRAS exon 3 (c59/c61) <NA>
## 3 BRAF exon 15 (c600) <NA> KRAS exon 3 (c59/c61) <NA>
## 4 BRAF exon 15 (c600) Wild-type KRAS exon 3 (c59/c61) Wild-type
## 5 BRAF exon 15 (c600) <NA> KRAS exon 3 (c59/c61) <NA>
## 6 BRAF exon 15 (c600) <NA> KRAS exon 3 (c59/c61) <NA>
## BMMTNM16 BMMTR16 BMMTR AE agecat61 BMI
## 1 NRAS exon 3 (c59/c61) <NA> Mutant Low old 24.60973
## 2 NRAS exon 3 (c59/c61) <NA> Mutant High young 20.70312
## 3 NRAS exon 3 (c59/c61) <NA> <NA> Low young 31.50110
## 4 NRAS exon 3 (c59/c61) Wild-type Mutant Low young 24.21875
## 5 NRAS exon 3 (c59/c61) <NA> <NA> Low old 20.31217
## 6 NRAS exon 3 (c59/c61) <NA> Mutant Low old 28.64736
## BMI_category
## 1 normal weight
## 2 normal weight
## 3 overweight/obese
## 4 normal weight
## 5 normal weight
## 6 normal weightfit.age <- survfit(Surv(PFSDYCR, PFSCR) ~ BMI_category, data=df_final1)
print(fit.age)## Call: survfit(formula = Surv(PFSDYCR, PFSCR) ~ BMI_category, data = df_final1)
##
## 1 observation deleted due to missingness
## n events median 0.95LCL 0.95UCL
## BMI_category=underweight 23 21 281 180 405
## BMI_category=normal weight 476 415 278 240 292
## BMI_category=overweight/obese 108 91 285 234 328survfit2(Surv(PFSDYCR, PFSCR) ~ BMI_category, data=df_final1) %>%
ggsurvfit(linewidth = 1) +
add_confidence_interval() + # add confidence interval
add_risktable() + # Add risk table
add_quantile(y_value = 0.5, color = "gray50", linewidth = 0.75)+ # Specify median survival
labs(title = "Kaplan-Meier Curve by BMI group",
x="Follow-up time (days)")+
scale_x_continuous(breaks = seq(0,1000,by=90))+
scale_color_discrete("BMI group")+
scale_fill_discrete("BMI group")+
add_pvalue("annotation",x=500)
# Add variables in a binary form
df_final1$race <- ifelse(df_final1$RACE %in% "White or Caucasian", "White or Caucasian", "Other")
df_final1$ECOG <- ifelse(df_final1$B_ECOG %in% c("Fully active", "Symptoms but ambulatory"), "0 or 1", "2")
df_final1$Primary_Tumor <- ifelse(df_final1$DIAGTYPE %in% "Colon", "Colon", "Rectal")
df_final1$Number_of_sites <- ifelse(df_final1$B_METANM == "1", "1",
ifelse(df_final1$B_METANM == "2", "2",
ifelse(df_final1$B_METANM >= 3, ">3", as.character(df_final1$B_METANM))))
df_final1$Location_of_site <- ifelse(df_final1$LIVERMET == "Y", "Liver", "Other")
colon2.1 <- subset(colon2, colon2$LBBASE == "Y" & colon2$LBTEST == "Lactate Dehydrogenase")
df_final1 <- merge(df_final1, colon2.1, by.x = "SUBJID" )
df_final1 <- df_final1[, !names(df_final1) %in% c("LBTEST", "LBBASE", "VISITDY", "VISIT", "LBSTUNIT")]
df_final1$Baseline_LDH1 <- ifelse(df_final1$LBSTRESN < 150 * 1.5, "< 1.5 x ULN",
ifelse(df_final1$LBSTRESN >= 150 * 1.5, ">= 1.5 x ULN", "other"))
df_final1$Baseline_LDH2 <- ifelse(df_final1$LBSTRESN < 200 * 2, "< 2 x ULN",
ifelse(df_final1$LBSTRESN >= 200 * 2, ">= 2 x ULN", "other"))
df_final1$Age1 <- ifelse(df_final1$AGE < 65, "<65", ">=65")
df_final1$Age2 <- ifelse(df_final1$AGE < 75, "<75", ">=75")# Caricare le librerie necessarie
library(survival)
library(dplyr)
library(forestplot)## Loading required package: grid## Loading required package: checkmate## Loading required package: abindlibrary(survminer)
# Creare un nuovo dataframe con tutte le variabili di interesse
df_final1 <- df_final1 %>%
mutate(AE = ifelse(AESEVCD %in% c(1, 2), "Low", "High"),
TRT = ifelse(TRT %in% "Panitumumab + FOLFOX", "Panitumumab + FOLFOX", "FOLFOX alone"),
race = ifelse(RACE %in% "White or Caucasian", "White or Caucasian", "Other"),
ECOG = ifelse(B_ECOG %in% c("Fully active", "Symptoms but ambulatory"), "0 or 1", "2"),
Primary_Tumor = ifelse(DIAGTYPE %in% "Colon", "Colon", "Rectal"),
Number_of_sites = ifelse(B_METANM == 1, "1",
ifelse(B_METANM == 2, "2",
ifelse(B_METANM >= 3, ">3", as.character(B_METANM)))),
Location_of_site = ifelse(LIVERMET == "Y", "Liver", "Other"),
Baseline_LDH1 = ifelse(LBSTRESN < 150 * 1.5, "< 1.5 x ULN",
ifelse(LBSTRESN >= 150 * 1.5, ">= 1.5 x ULN", "other")),
Baseline_LDH2 = ifelse(LBSTRESN < 200 * 2, "< 2 x ULN",
ifelse(LBSTRESN >= 200 * 2, ">= 2 x ULN", "other")),
Age1 = ifelse(AGE < 65, "<65", ">=65"),
Age2 = ifelse(AGE < 75, "<75", ">=75"))
# Modello Cox PH con tutte le variabili
cox_model_full <- coxph(Surv(PFSDYCR, PFSCR) ~ AE + TRT + BMMTR + ECOG + Primary_Tumor + Number_of_sites +
Location_of_site + Baseline_LDH1 + Baseline_LDH2 + Age1 +
Age2 + SEX + race + PRSURG + BMI_category, data = df_final1)
# Sommario del modello
summary_cox_full <- summary(cox_model_full)
coef(cox_model_full)## AELow TRTPanitumumab + FOLFOX
## 0.204926907 0.097100453
## BMMTRWild-type ECOG2
## -0.533819479 0.638106125
## Primary_TumorRectal Number_of_sites1
## -0.070345847 -0.172415606
## Number_of_sites2 Location_of_siteOther
## -0.128567807 0.165999001
## Baseline_LDH1>= 1.5 x ULN Baseline_LDH2>= 2 x ULN
## 0.278796683 0.017285654
## Age1>=65 Age2>=75
## 0.091164105 0.101553984
## SEXMale raceWhite or Caucasian
## -0.008261972 0.065201933
## PRSURGY BMI_categorynormal weight
## -0.305881121 0.120761565
## BMI_categoryoverweight/obese
## 0.018181541# Estrarre i risultati
HR <- exp(coef(cox_model_full)) # Hazard Ratios
CI_lower <- exp(confint(cox_model_full)[, 1]) # Limiti inferiori degli intervalli di confidenza
CI_upper <- exp(confint(cox_model_full)[, 2]) # Limiti superiori degli intervalli di confidenza
p_values <- summary_cox_full$coefficients[, 5] # P-values
# Preparazione dei dati per il forest plot
forest_data <- data.frame(
Factors = c("Adverse Events Grade: Low", "Treatment: Panitumumab + FOLFOX", "KRAS: WT", "ECOG: 2 (in bed < 50% of the time)", "Primary Tumor: Rectal", "Number of Sites: 1", "Number of Sites: 2", "Location of Site: Other", "Baseline LDH1: >= 1.5 x ULN", "Baseline LDH2: >= 2 x ULN", "Age: >=65", "Age: >=75", "Sex: Male", "Race: White or Caucasian", "Pre-Surgery", "BMI category: normal weight", "BMI category: overweight/obese"),
HR = round(HR, 2),
Lower_CI = round(CI_lower, 2),
Upper_CI = round(CI_upper, 2),
p_value = round(p_values, 3)
)
# Preparare i testi da mostrare nella colonna di etichette
labels <- cbind(
forest_data$Factors,
sprintf("%.2f", forest_data$HR),
sprintf("%.2f - %.2f", forest_data$Lower_CI, forest_data$Upper_CI)
)
# Creare il forest plot
forestplot(
labeltext = labels,
mean = forest_data$HR,
lower = forest_data$Lower_CI,
upper = forest_data$Upper_CI,
title = "Hazard Ratio by Factor",
xlab = "Hazard Ratio (log scale)",
is.summary = rep(FALSE, nrow(forest_data)),
xlog = TRUE,
xticks = c(0.5, 1, 2),
col = forestplot::fpColors(box = "blue", line = "darkblue", summary = "royalblue"),
lineheight = unit(1.5, "cm"), # Aumenta l'altezza delle linee per spazio tra le etichette
boxsize = 0.2, # Riduci la dimensione dei box per una visualizzazione più pulita
clip = c(0.5, 2.5), # Limita l'asse x per una visualizzazione più chiara
txt_gp = fpTxtGp(
label = gpar(fontsize = 10), # Aumenta la dimensione del testo per le etichette
xlab = gpar(fontsize = 12), # Aumenta la dimensione del testo per l'asse x
title = gpar(fontsize = 14, fontface = "bold") # Aumenta la dimensione del titolo
)
)
library(survival)
library(survminer)
library(cowplot)##
## Attaching package: 'cowplot'## The following object is masked from 'package:ggpubr':
##
## get_legend# Fit the Cox PH model
cox_model <- coxph(Surv(PFSDYCR, PFSCR) ~ AE + TRT + BMMTR + ECOG + Primary_Tumor + Number_of_sites +
Location_of_site + Baseline_LDH1 + Baseline_LDH2 + Age1 +
Age2 + SEX + race + PRSURG + BMI_category, data = df_final1)
# Schoenfeld Residuals Test
cox_zph <- cox.zph(cox_model)
# Print the test results
print(cox_zph)## chisq df p
## AE 0.434 1 0.51
## TRT 0.204 1 0.65
## BMMTR 0.481 1 0.49
## ECOG 2.312 1 0.13
## Primary_Tumor 1.490 1 0.22
## Number_of_sites 1.494 2 0.47
## Location_of_site 1.018 1 0.31
## Baseline_LDH1 0.917 1 0.34
## Baseline_LDH2 1.063 1 0.30
## Age1 0.175 1 0.68
## Age2 0.174 1 0.68
## SEX 0.105 1 0.75
## race 0.701 1 0.40
## PRSURG 0.261 1 0.61
## BMI_category 0.322 2 0.85
## GLOBAL 10.417 17 0.89# Plot diagnostics
# Schoenfeld Residuals Test
cox_zph <- cox.zph(cox_model_full)
# Generate diagnostic plots
ggcoxzph(cox_zph)
plot_dfbeta <- ggcoxdiagnostics(cox_model_full, type = "dfbeta", linear.predictions = FALSE, ggtheme = theme_bw())## Warning: `gather_()` was deprecated in tidyr 1.2.0.
## ℹ Please use `gather()` instead.
## ℹ The deprecated feature was likely used in the survminer package.
## Please report the issue at <https://github.com/kassambara/survminer/issues>.
## This warning is displayed once every 8 hours.
## Call `lifecycle::last_lifecycle_warnings()` to see where this warning was
## generated.plot_deviance <- ggcoxdiagnostics(cox_model_full, type = "deviance", linear.predictions = FALSE, ggtheme = theme_bw())
plot_martingale <- ggcoxdiagnostics(cox_model_full, type = "martingale", linear.predictions = FALSE, ggtheme = theme_bw())
plot_dfbeta## `geom_smooth()` using formula = 'y ~ x'
# Combine the plots into a single grid
combined_plot <- plot_grid(plot_deviance, plot_martingale, labels = "AUTO")## `geom_smooth()` using formula = 'y ~ x'## `geom_smooth()` using formula = 'y ~ x'# Display the combined plot
print(combined_plot)
library(tidyverse)## ── Attaching core tidyverse packages ──────────────────────── tidyverse 2.0.0 ──
## ✔ forcats 1.0.0 ✔ stringr 1.5.1
## ✔ lubridate 1.9.3 ✔ tibble 3.2.1
## ✔ purrr 1.0.2 ✔ tidyr 1.3.1
## ✔ readr 2.1.5
## ── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
## ✖ gridExtra::combine() masks dplyr::combine()
## ✖ tidyr::extract() masks magrittr::extract()
## ✖ dplyr::filter() masks stats::filter()
## ✖ dplyr::lag() masks stats::lag()
## ✖ purrr::set_names() masks magrittr::set_names()
## ✖ lubridate::stamp() masks cowplot::stamp()
## ℹ Use the conflicted package (<http://conflicted.r-lib.org/>) to force all conflicts to become errors# Filtrare solo i dati di screening
colon_screening <- colon2 %>% filter(VISIT == "Screening")
# Riorganizzare i dati con pivot_wider
colon_wide <- colon_screening %>%
select(SUBJID, LBTEST, LBSTRESN) %>%
pivot_wider(names_from = LBTEST, values_from = LBSTRESN)## Warning: Values from `LBSTRESN` are not uniquely identified; output will contain
## list-cols.
## • Use `values_fn = list` to suppress this warning.
## • Use `values_fn = {summary_fun}` to summarise duplicates.
## • Use the following dplyr code to identify duplicates.
## {data} |>
## dplyr::summarise(n = dplyr::n(), .by = c(SUBJID, LBTEST)) |>
## dplyr::filter(n > 1L)# Convertire i dati in numerico
colon_wide <- colon_wide %>%
mutate(across(everything(), ~ as.numeric(as.character(.))))## Warning: There were 8 warnings in `mutate()`.
## The first warning was:
## ℹ In argument: `across(everything(), ~as.numeric(as.character(.)))`.
## Caused by warning:
## ! NAs introduced by coercion
## ℹ Run `dplyr::last_dplyr_warnings()` to see the 7 remaining warnings.# Rimuovere le righe con NA
colon_wide <- colon_wide %>% drop_na()
# Identificare i duplicati
colon_wide <- colon_wide %>% distinct(SUBJID, .keep_all = TRUE)
# Controllare i dati riorganizzati
print(colon_wide)## # A tibble: 816 × 9
## SUBJID Albumin `Alkaline Phosphatase` Creatinine `Lactate Dehydrogenase`
## <dbl> <dbl> <dbl> <dbl> <dbl>
## 1 3 3.55e 1 314 9.99e15 271
## 2 4 3.4 e 1 475 6.8 e 1 666
## 3 5 3.7 e 1 122 6.19e 1 381
## 4 6 2.88e 1 100 7.25e 4 249
## 5 7 3.4 e 1 253 7.07e 1 309
## 6 9 5.1 e 1 110 6.19e 1 214
## 7 10 3.8 e 1 313 8.04e 4 749
## 8 13 4.19e 1 308 4.42e 1 308
## 9 15 4.4 e 1 61 7.07e 1 280
## 10 16 3.78e16 365 9.28e15 1131
## # ℹ 806 more rows
## # ℹ 4 more variables: Hemoglobin <dbl>, Platelets <dbl>,
## # `White Blood Cells` <dbl>, `Carcinoembryonic Antigen` <dbl># Unire il dataframe riorganizzato con il dataframe principale
df_final1 <- merge(df_final1, colon_wide, by = "SUBJID")
# Controllare i dati uniti
print(head(df_final1))## SUBJID AEPT AESOC AESTDYI
## 1 3 Dermatitis acneiform Skin and subcutaneous tissue disorders 12
## 2 7 Folliculitis Infections and infestations 245
## 3 9 Exfoliative rash Skin and subcutaneous tissue disorders 3
## 4 13 Rash Skin and subcutaneous tissue disorders 37
## 5 15 Dry skin Skin and subcutaneous tissue disorders 72
## 6 17 Urticaria Skin and subcutaneous tissue disorders 175
## AEENDYI AESEVCD AECONT TRT ATRT PRSURG DTHDY
## 1 29 1 N Panitumumab + FOLFOX Panitumumab + FOLFOX Y 301
## 2 252 3 N Panitumumab + FOLFOX Panitumumab + FOLFOX Y 339
## 3 14 2 N Panitumumab + FOLFOX Panitumumab + FOLFOX Y 696
## 4 55 2 N Panitumumab + FOLFOX Panitumumab + FOLFOX Y 197
## 5 157 2 N Panitumumab + FOLFOX Panitumumab + FOLFOX Y 414
## 6 175 2 N FOLFOX alone FOLFOX alone Y 1043
## DTH PFSDYCR PFSCR LIVERMET DIAGMONS AGE SEX B_WEIGHT B_HEIGHT
## 1 1 54 1 Y 1.86885245901639E+16 65 Male 67 165
## 2 1 339 1 Y 2.78688524590163E+16 52 Female 53 160
## 3 1 301 1 Y 7570491803278680 42 Female 69 148
## 4 1 73 1 Y 2.0327868852459E+16 54 Male 62 160
## 5 1 414 1 Y 380327868852459 67 Female 48.8 155
## 6 0 226 0 Y 3.14754098360655E+16 57 Female 63.6 149
## RACE B_ECOG HISSUBTY B_METANM DIAGTYPE
## 1 White or Caucasian Symptoms but ambulatory No sub-type 5 Colon
## 2 Hispanic or Latino Symptoms but ambulatory Mucinous 3 Colon
## 3 Hispanic or Latino Fully active No sub-type 4 Colon
## 4 White or Caucasian Fully active Other 6 Rectal
## 5 White or Caucasian Symptoms but ambulatory No sub-type 2 Rectal
## 6 White or Caucasian Fully active No sub-type 2 Colon
## BMMTNM1 BMMTR1 BMMTNM2 BMMTR2
## 1 KRAS exon 2 (c12/13) Mutant KRAS exon 3 (c61) <NA>
## 2 KRAS exon 2 (c12/13) Mutant KRAS exon 3 (c61) <NA>
## 3 KRAS exon 2 (c12/13) Failure KRAS exon 3 (c61) <NA>
## 4 KRAS exon 2 (c12/13) Wild-type KRAS exon 3 (c61) Wild-type
## 5 KRAS exon 2 (c12/13) Failure KRAS exon 3 (c61) <NA>
## 6 KRAS exon 2 (c12/13) Mutant KRAS exon 3 (c61) <NA>
## BMMTNM3 BMMTR3 BMMTNM4 BMMTR4
## 1 KRAS exon 4 (c117/146) <NA> NRAS exon 2 (c12/13) <NA>
## 2 KRAS exon 4 (c117/146) <NA> NRAS exon 2 (c12/13) <NA>
## 3 KRAS exon 4 (c117/146) <NA> NRAS exon 2 (c12/13) <NA>
## 4 KRAS exon 4 (c117/146) Mutant NRAS exon 2 (c12/13) Wild-type
## 5 KRAS exon 4 (c117/146) <NA> NRAS exon 2 (c12/13) <NA>
## 6 KRAS exon 4 (c117/146) <NA> NRAS exon 2 (c12/13) <NA>
## BMMTNM5 BMMTR5 BMMTNM6 BMMTR6
## 1 NRAS exon 3 (c61) <NA> NRAS exon 4 (c117/146) <NA>
## 2 NRAS exon 3 (c61) <NA> NRAS exon 4 (c117/146) <NA>
## 3 NRAS exon 3 (c61) <NA> NRAS exon 4 (c117/146) <NA>
## 4 NRAS exon 3 (c61) Wild-type NRAS exon 4 (c117/146) Wild-type
## 5 NRAS exon 3 (c61) <NA> NRAS exon 4 (c117/146) <NA>
## 6 NRAS exon 3 (c61) <NA> NRAS exon 4 (c117/146) <NA>
## BMMTNM7 BMMTR7 BMMTNM15 BMMTR15
## 1 BRAF exon 15 (c600) <NA> KRAS exon 3 (c59/c61) <NA>
## 2 BRAF exon 15 (c600) <NA> KRAS exon 3 (c59/c61) <NA>
## 3 BRAF exon 15 (c600) <NA> KRAS exon 3 (c59/c61) <NA>
## 4 BRAF exon 15 (c600) Wild-type KRAS exon 3 (c59/c61) Wild-type
## 5 BRAF exon 15 (c600) <NA> KRAS exon 3 (c59/c61) <NA>
## 6 BRAF exon 15 (c600) <NA> KRAS exon 3 (c59/c61) <NA>
## BMMTNM16 BMMTR16 BMMTR AE agecat61 BMI
## 1 NRAS exon 3 (c59/c61) <NA> Mutant Low old 24.60973
## 2 NRAS exon 3 (c59/c61) <NA> Mutant High young 20.70312
## 3 NRAS exon 3 (c59/c61) <NA> <NA> Low young 31.50110
## 4 NRAS exon 3 (c59/c61) Wild-type Mutant Low young 24.21875
## 5 NRAS exon 3 (c59/c61) <NA> <NA> Low old 20.31217
## 6 NRAS exon 3 (c59/c61) <NA> Mutant Low old 28.64736
## BMI_category race ECOG Primary_Tumor Number_of_sites
## 1 normal weight White or Caucasian 0 or 1 Colon >3
## 2 normal weight Other 0 or 1 Colon >3
## 3 overweight/obese Other 0 or 1 Colon >3
## 4 normal weight White or Caucasian 0 or 1 Rectal >3
## 5 normal weight White or Caucasian 0 or 1 Rectal 2
## 6 normal weight White or Caucasian 0 or 1 Colon 2
## Location_of_site LBSTRESN Baseline_LDH1 Baseline_LDH2 Age1 Age2 Albumin
## 1 Liver 271 >= 1.5 x ULN < 2 x ULN >=65 <75 35.5
## 2 Liver 309 >= 1.5 x ULN < 2 x ULN <65 <75 34.0
## 3 Liver 214 < 1.5 x ULN < 2 x ULN <65 <75 51.0
## 4 Liver 308 >= 1.5 x ULN < 2 x ULN <65 <75 41.9
## 5 Liver 280 >= 1.5 x ULN < 2 x ULN >=65 <75 44.0
## 6 Liver 93 >= 1.5 x ULN >= 2 x ULN <65 <75 38.0
## Alkaline Phosphatase Creatinine Lactate Dehydrogenase Hemoglobin Platelets
## 1 314 9.9892e+15 271 103 279
## 2 253 7.0720e+01 309 100 527
## 3 110 6.1880e+01 214 144 294
## 4 308 4.4200e+01 308 124 625
## 5 61 7.0720e+01 280 105 296
## 6 105 5.7000e+01 85 128 284
## White Blood Cells Carcinoembryonic Antigen
## 1 5.600e+16 7.200e+00
## 2 7.100e+16 1.210e+02
## 3 8.600e+00 4.166e+16
## 4 1.162e+16 5.000e-01
## 5 9.400e+00 8.728e+01
## 6 8.100e+00 2.900e+01df_final1$AE <- ifelse(df_final1$AE == "Low", 0, 1)
# Fit the logistic regression model
logistic_model <- glm(AE ~ `Albumin` + `Alkaline Phosphatase` + `Creatinine` +
`Lactate Dehydrogenase` + `Hemoglobin` + `Platelets` +
`White Blood Cells` + `Carcinoembryonic Antigen` + TRT,
family = binomial, data = df_final1)
# Summarize the model
summary(logistic_model)##
## Call:
## glm(formula = AE ~ Albumin + `Alkaline Phosphatase` + Creatinine +
## `Lactate Dehydrogenase` + Hemoglobin + Platelets + `White Blood Cells` +
## `Carcinoembryonic Antigen` + TRT, family = binomial, data = df_final1)
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -2.640e+00 3.488e-01 -7.568 3.79e-14 ***
## Albumin 5.356e-18 1.358e-17 0.394 0.6933
## `Alkaline Phosphatase` -6.481e-17 7.937e-17 -0.817 0.4142
## Creatinine 2.751e-17 1.585e-17 1.736 0.0826 .
## `Lactate Dehydrogenase` -2.832e-17 6.044e-17 -0.469 0.6394
## Hemoglobin 2.154e-17 8.691e-17 0.248 0.8043
## Platelets -6.721e-16 3.213e-14 -0.021 0.9833
## `White Blood Cells` -2.533e-18 5.609e-18 -0.452 0.6516
## `Carcinoembryonic Antigen` -2.858e-18 1.058e-17 -0.270 0.7871
## TRTPanitumumab + FOLFOX 2.050e+00 3.625e-01 5.656 1.55e-08 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 635.38 on 532 degrees of freedom
## Residual deviance: 577.61 on 523 degrees of freedom
## AIC: 597.61
##
## Number of Fisher Scoring iterations: 14